There is a clamor and hype surrounding AI. The speed at which the tech is developing has raised some ethical concerns. Among this voice, the CEO of Bitmix has a different opinion.
There has been a single way tech buildout rolls out:
Tech Rollout Overbuilds ➔ Market Crash / Failures ➔ Government Bailout (Money Printing) ➔ Currency Devaluation ➔ Capital Flees to Crypto
As per Arthur Hayes’ recent statement, this is how investment in AI infra will actually end. This seems like the most probable scenario.
In recent days, conversations around AI have taken a cautious turn- from photo-ops with the President to a new label, “superintelligence.” This marketing play has made the masses fairly aware, and more skeptical, of AI’s actual growth than ever before.
But there’s not much that can be done when those with the power to make the move want to see it through as the first of its kind, the most innovative breakthrough in any field. This is alongside the various “performative” warnings (accompanied by nods) and the need for regulations delegated by the AI Avengers- from Musk to Amodei.
And if that wasn’t any simpler- Huang disagreed with the need for regulations on AI models, and Sam Altman basically told the world to ‘suck it up’: what are a few grievances in the face of innovation? Imagine superintelligence.
However, the BitMEX CEO might be on the right side of the fence. He believes humanity (the tech bros) investing trillions of dollars into AI data centers is a sheer waste of time. The massive data center buildout is going nowhere, and as with any other tech rollout (such as the 19th-century railways or the 1990s Dot-Com fiber-optic buildout), it will end in an asset crash.
With AI infra being built at such a scale, wouldn’t computing power become cheap and accessible? It might; that’s the unsure part. But the pressure point is that these AI and tech companies are burning cash- without matching bottom-line impact.
Is overbuilding infrastructure the right thing to do when it’s leading to bailouts but not profit?
This creates a setback. And that has been the problem with AI.
The tech itself (aside from generative AI) hasn’t been a bottleneck. There are fields where AI’s potential is quite evident. It’s the over-investment, the promises, and the over-hyping that’s worrying skeptics.
“Once the data centers currently under construction are completed, infrastructure providers will seek payment for the compute those companies have committed to,” emphasizes Arthur Hayes.
